Enrique Blanco

The right way to read MOSS user and lookup field values

Enrique Blanco — Sat, 17 Oct 2009 17:17:01 GMT

Sometimes when we read the value of certain fields (lookup, user, links, etc), the value of the fields is kind of “encoded”:

For example for the FileRef field can be something like “1;#Paginas/default.aspx”

A weird value ain’t it?, what people usually do is split by # and then get the second part, that is a quite sad method…

How to properly read the values of those encoded fields?

First of all you must know which type the value corresponds to, so open your SharePoint Manager (if you don’t use any tool like it, you should) and get to the list and field in question:

You can see the internal name of the field (in this case FileRef) and in the FieldValueType property the name of the class you should use to get the values of the field (in this case SPFieldLookupValue), so all you have to do to set the value of the field is using the class to parse the values and use the members.

SPFieldLookupValue spFValue = new SPFieldLookupValue(itemPage[“FileRef”].ToString());
string url = spFValue.LookupValue;

For the Created By field, you should use the SPFieldUserValue definition, for the PublishingPageLayout use the SPFieldUrlValue… everything can be achieved by inspecting the list definitions with SPM, so you have no excuses ;).

The main point against using the Split method is that the storage method can change with a service pack or in next versions of the product and then your code might fail at runtime, if there is a method for accesing that fields properly we should use that.

FullTextSqlQuery, SPContext.Current.Site and DISP_E_Exception

Enrique Blanco — Sun, 27 Sep 2009 13:45:05 GMT

I had quite a weird error, when deleting a webpart I got an error like this, it was obvious that some object was already disposed but I had made sure (or at least I thought) that I had all the memory disposal issues under control.

After commenting various functions I had targeted the error to a code region like this:

using (FullTextSqlQuery query = new FullTextSqlQuery(SPContext.Current.Site))
{ … }

And after looking in msdn there was a community comment that said:

“You need to be very careful about what SPSite object you pass to this constructor. If you pass in the SPContext.Current.Site
object and you later call the Dispose() method on your instance of the FullTextSqlQuery object (or if you've wrapped the instance in a using statement), you can potentially get the dreaded "Trying to use and SPWeb object that has been closed or disposed ..." error later in the page request. That's because the FullTextSqlQuery class keeps a reference of the SPSite object instance passed into the constructor and when this FullTextSqlQuery instance is disposed, the SPSite object is also disposed”

So the code (for now will be):

Guid idSitio = SPContext.Current.Site.ID;
using (SPSite sitio = new SPSite(idSitio))
{
using (FullTextSqlQuery query = new FullTextSqlQuery(sitio))
{ … }

}

Thanks to Bart X Tubalinal ;)

Inconsistent checked out state in SPD - Clean SharePoint Designer cache

Enrique Blanco — Sun, 27 Sep 2009 12:53:42 GMT

This is a classic one, but quite annoying…

You’re editing a file (a css for example) in SPD and it tells you that the file is checked out, but you know it isn’t. You try to check it in to get things in sync and it raises an error, because the file is not really checked out… you try F5, exiting and entering SPD and cleaning IE cache and nothing works…

The solution is exiting SPD, opening C:\Documents and Settings\<user>\Configuración local\Datos de programa\Microsoft\WebsiteCache (or similar in your language) and deleting the folder of the site you’re trying to connect to (I will delete all files and folder there).

Once deleted you open SPD and the view you get is in sync with the correct state of the files in the server.

I think my problem is that I had the css checked out and then deleted the site collection and recreated it, so SPD cache was invalid, what I don’t know is why there is not a visible “Refresh all” button, because F5 did not worked for me this time…

SQL Server Full Text Search + iFilters + documents in blobs

Enrique Blanco — Tue, 22 Sep 2009 23:38:00 GMT

A usually underused feature of SQL Server is the Full Text Search, usually we simply code LIKE statements and do not create catalogs to have more advances queries.

A very interesting feature of the Full Text Search engine is the integration with the iFilters that allow you not only search in the text fields, but also in the image (blob) fields that can contain pdf or doc files, so your application can perform also documental search over the files, without having to use WSS to store the documents or index them. Much simpler.

The only “bad” point is that you need to install the iFilters on the SQL Servers, and some administrators don’t like the idea to touch them too much.

You can download the iFilter Pack for free from MSDN. The installation is quite straightforward and does not require to reboot the machine. After installing the binaries you need to register the iFilters in SQL, you have the instructions for SQL Server 2008 here.

Then, for the test, we will create a new database called, for example, Test_DB.

I will create a new table with an ID, an image Field and a varchar field:

CREATE TABLE [dbo].[Test_Table](
    [ID] [uniqueidentifier] NOT NULL,
    [ImageField] [varbinary](max) NOT NULL,
    [Extension] [varchar](50) NOT NULL,
CONSTRAINT [PK_Test_Table] PRIMARY KEY CLUSTERED
(
    [ID] ASC
)

The ID is the primary key field, the Image field will host the docs, and the extension field will get the extension of the files, so SqlServer can call the proper iFilter to process the file.

Then we will create a new Full Text Catalog, in the management studio, selecting the database and under the storage section, Full Text Catalogs:

Once you create the full text catalog you should make it the default catalog: Right click over the catalog-> Properties-> Default Catalog:

Then we will instruct SQL Server to index the column using the appropriate iFilter:

CREATE FULLTEXT INDEX ON Test_Table
(ImageField TYPE COLUMN Extension LANGUAGE 'SPANISH')
KEY INDEX PK_Test_Table

We tell in this command to create the index on Test_Table, to index the ImageField field, use the column “Extension” to get the extension of the file (and the iFilter to use), the language used for partitioning the words and the “PK_Test_Table” is the name of the key created for the Primary Key of the table:

Then we will add a couple of documents to the table:

Then we rebuild the catalog, for SQL to index it. We can also set a population schedule to keep the catalogs updated with the DB.

After that when we see the properties of the catalog we can see that the two docs are indexed:

Then all we have to do is search by a word contained in the documents, and magically you get the results!

I think this is a very useful technique to give an added value to your applications.

Finding HRESULT: 0x80020009 (DISP_E_EXCEPTION)

Enrique Blanco — Wed, 16 Sep 2009 20:48:00 GMT

Sometimes you get this error, and (as usually) the log does not provide a lot of information, the error seems to be caused in MOSS own source code and apparently has nothing to do with yours. The error really says that an already disposed object is trying to be used.

When will you usually get this exception? When you assign an already created object to a variable, dispose the variable and try to use the first object again, the most typical case is something like this:

using (SPSite site = SPContext.Current.Site)

{

…

}

The using disposes automatically the “site” variable and when the system (or yourself) tries to use the SPContext.Current.Site object the exception arises from the system code, so it can be difficult to debug and find the source of the problem.

Advanced CSS eBook

Enrique Blanco — Mon, 14 Sep 2009 22:11:47 GMT

Via twitter my friend Bruno pointed us to this site, that contains this excellent ebook about advanced css.

I’ve had only time to take a quick look at it, but it seems very complete and easy to follow.

How to add a new server to an existing upgraded farm

Enrique Blanco — Sun, 02 Aug 2009 09:27:52 GMT

This is a very simple one, imagine you have an existing farm, all the servers have been upgraded: SP1, language packs, cumulative updates… and you have to add an additional server to the farm. What is the sequence of steps?

Here it is:

First of all, you can create an installation disk that includes the updates (sp1 or sp2), specially useful when the farm is heavy “patched”, you can see the procedure right here: http://technet.microsoft.com/en-us/library/cc287882.aspx

You must get the version of moss you are currently running, you can see how here: http://blogs.renacimiento.com/aarias/archive/2009/06/16/[moss]versión-de-sharepoint-instalada.aspxç

And you must know which language packs are you running, you can get that by looking at the available language options when creating a new site collection, or by looking for the typical 1033, 3082, etc folders in the 12 hive directories. You have a references for the country codes in http://technet.microsoft.com/en-us/library/cc262108.aspx.

Then you have to install the software in the proper order, WITHOUT RUNNING THE SHAREPOINT PRODUCTS AND TECHNOLOGIES WIZARD:

Moss

Wss Language Pack

Moss Language Pack

Moss SP1/SP2

Wss Language Pack SP1/SP2

Moss Language Pack SP1/SP2

Post SP1/SP2 cumulative updates if needed.

SP1/SP2 means one or another, preferably SP2 (in my case the customer did not wanted to upgrade to SP2 so soon).

While you installing some of this software the Products and Technologies Wizard will try to run, you simply have to cancel it.

Once you have installed all the software you have to add the server to the existing farm by running the Products and Technologies wizard on the server and selecting to join an existing farm, then it will ask you for the SQL server instance your farm configuration database is, once selected the new server will configure and will be a part of the farm.

NOTE: If you decide not to upgrade to SP2, you must not install any SP2 labeled files (that is the WSS Language Pack SP2), because it will alter some of your dll’s version numbers and problems will arise when installing some patches.

The proper way to handle the Name ActiveX Control warning

Enrique Blanco — Sat, 01 Aug 2009 19:26:07 GMT

One common mistake in many public facing MOSS/WSS sites (including this blog) is the Name ActiveX Control warning, it seems its the presence control activeX, that can be very useful in an intranet environment, but its no good at all in public facing sites:

MS gives us a solution http://support.microsoft.com/kb/931509, but it is a really crappy one, you can add the site to the trusted sites or you can copy the init.js and reference the copy in your design… then you’ll might have to take it into account when a service pack is applied and things so..

I’ll tell you how I made it in my current project: A public facing site that uses webpartzones, and must be AA compatible.

Removing the default scripts when not needed

First of all, the only moment when I need MOSS’s default scripts and css is when an editor is working on the page, not when the visitors are in the site, so we will only show all the MOSS scripts and css when the user in an editor, we will archieve that by inserting the references inside an AuthoringContainer that is showed only when the user is an author:

After that you’ll see that when an anonymous user visits the page, MOSS does not render the script and css references, the page is lighter, loads faster and the ActiveX warning is gone, but if you use webparts, the script generated by the SPWebPartManager causes a javascript error at the end of the page in this script:

If you think about it, for anonymous users there is no need for this script since they are not editing the page, moving or exporting webparts, so… let’s fix it.

We have two options:

Introducing a fake script to do nothing:
Removing completely the script and thus reducing the page size.

Fake Script

It’s a simple but IMHO a bit incomplete solution, it consists of setting up a fake script that makes nothing so the browser does not raise a javascript error, you can get the script here:

http://social.msdn.microsoft.com/Forums/en-US/sharepointcustomization/thread/46ecb6c0-ad30-4e8f-8e46-e3b95bde9b6a

You can put this script on a container that only shows for anonymous users, and the browser will not cry. ;)

Remove the script

The fake script solution worked for me, but I do not want all those unused, heavy scripts on my page, so I need a way to eliminate them, so after a bit of “Reflectoring” I saw that the sealed SPWebpartManager used the RegisterClientScript functions to emit that script into the page.

Both the ClientScriptManager and the SPWebPartManager were sealed classes (why in the hell does MS seal all interesting classes?), so I couldn’t inherit them to control when to emit the offending javascript and hidden fields, so I had to use reflection to simply “clean” the script before rendering the page:

I named the class “ScriptManagerFixer” and made some functions to eliminate the generated scripts. You can download the class here.

So I need a point from where to call the functions, so I chose my MasterPage’s Render method:

You can see that I eliminate all the scripts generated by the SPWebPartManager class and its base class, WebPartManager, I also eliminate all the hidden fields, so I can save some kbs in the html size.

With this solution you’ll get the full package, you’ll eliminate the script error and also will make your pages lighter.

Hope it helps!!

Exporting Scopes, Crawled and Managed Properties between SSPs

Enrique Blanco — Fri, 03 Jul 2009 22:50:00 GMT

Imagine this, you are a happy developer that is creating a wonderful enterprise search webpart for MOSS, you have created lots and lots of scopes and managed properties in your machine and now you have to install your app in the test, QA and production server and your team also need the app in their machines, you need to recreate the scopes and properties in the machines… This can be a real hell.

If you are in this situation SharePoint Shared Services Search Provider Property Creation is your tool.

Simply export and import, and all the machines will be alike :), really really time saving!

Troubleshooting MOSS Explorer View

Enrique Blanco — Sun, 21 Jun 2009 10:40:15 GMT

Just a minutes ago I came across this whitepaper: Understanding and Troubleshooting the SharePoint Explorer View

Explorer view is one of the features that causes more user complains, the main point about it is that needs the Web Client service running on the client machine and it is not always running, so it might cause trouble.

I’ve seen also problems with IE8 not displaying it, in my VM setup (Windows 2003), I managed to get it running by:

Making sure WebDav is enabled in IIS.
Making sure WebClient service is running in the client machine. Please make sure you run windows update on the clients, this service has suffered from some important vulnerabilities.
Upgrading WSS and MOSS to Service Pack 2.

Hope it helps.